Anthony Giddens:  Structure and Function

John Hamilton Bradford, Ph.D.

Giddens and the (failed) attempt to transcend the agency-system distinction

There have been various attempts to “transcend” this distinction, the most famous of which has been Giddens’ theory of structuration.  Giddens is interested in how social systems, defined as reproduced relations between actors as organized social practices, are recursively reconstituted as the interaction of intended and unintended consequences of action.  Specifically, he notes how the unintended consequences of action “may systematically feed back to be the unacknowledged conditions of further acts” (1984: 8).  Social systems are “both medium and outcome of the practices they recursively organize” (25: my emphasis).

For Giddens, structures consist of “rules and resources” and exist outside of time and space and are hence virtual. They exist rather as memory traces.  Giddens writes that structure is not external, but internal to individuals.  Individuals are constrained and enabled in their actions, that is to say, individuals have capacities to act but cannot act in any way they can imagine.  Not every action is possible.  The repeated thesis of his structuration theory is that is not exclusively constraining, as is implied by the dualism of society versus the individual. Instead, structure is seen as a duality, which is both enabling and constraining. The duality of structure refers to the mutual constitution of agents and structures:

The constitution of agents and structures are not two independently given sets of phenomena, a dualism, but represent a duality. According to the notion of the duality of structure, the structural properties of social systems are both medium and outcome of the practices they recursively organize. Structure is not 'external' to individuals: as memory traces, and as instantiated in social practices, it is in a certain sense more 'internal' than exterior to their activities in a Durkheimian sense. Structure is not to be equated with constraint but is always both constraining and enabling. (1984: 25)

Whereas structures exist virtually, that is, are only implied (“implicated”), systems are the patterned social practices of human agents distributed across space and time.  From this perspective, structures do not “structure” action in the sense of determining the pattern or the system-ness that the system exhibits.  Structure enables and constrains, but does not determine, the system, understood as the aggregated outcomes of actions.[1]

Patterns of social interaction arise from micro-level agencies which are latent from the perspective of the actors themselves. As Giddens observes: “human history is created by intentional activities but is not an intended project; it persistently eludes efforts to bring it under conscious direction” (1984: 279). The idea is similar to Marx's notion that: “Men make their own history, but they do not make it just as they please; they do not make it under circumstances chosen by themselves, but under circumstances directly encountered, given and transmitted from the past.”

Giddens writes that social systems require human agency but that “it is not the case that actors create social systems; they reproduce or transform them, remaking what is already made in continuity of praxis” (1984: 171).  Social systems do not arise ex nihilo, and so the question of how it all began (i.e. the question of the diachronic emergence of social systems) is not addressed.  Aggregated actions constitute systems or patterns of interaction that are latent for the individual actors themselves. These patterns should be seen as the interaction of both intended and unintended consequences of actions.

Giddens claims that his theory overcomes subject-object dualism because for him, structure is not external to the actor.  Structure is both ‘medium and outcome of the conduct it recursively organizes’.   As medium, structure furnishes rules and resources that enable and make social life possible; as outcome, its reproduction result in the instantiation of rules in action and interaction (i.e. system).  

As Nicos Mouzelis (1995) observers, the duality-of-structure thesis implies a particular type of subject-object relationship that subjects adopt towards rules and resources that can be characterized as natural-performative.  The natural-performative orientation to rules and resources is a practical, natural, taken for granted attitude.  However, this view omits another orientation that can be described as strategic-monitoring of rules and resources .  The same point about social structure can be made with respect to the actual rules and resources instantiated by actors, i.e. to social systems.  

Table 6.  Mouzelis and the points of view one can adopt towards rules and resources.

This results in a fourfold typology specifying four broad perspectives or orientations can actors can adopt vis-a-vis rules and resources, depicted in Table 6.  

For Giddens, the agency-structure duality refers to cases in which structure is inside the individual.  However, for cases in which individuals belong to much larger, abstract systems in which he/she only participates minimally, the concept of a dualism(which Giddens rejects) is more appropriate.  Mouzelis therefore distinguishes between a syntagmatic duality (ala Giddens) and a syntagmatic dualism.  The term syntagmatic refers to institutional, or structural (as Giddens uses the term) rules and resources.  The distinction bewteen dualism and duality is again recapiulated at the level of historically instantiated systems.  Mouzelis therefore distinguishes between paradigmatic dualism and paradigmatic duality.  

The point is that the distinction is just that, a distinction, which is more over useful for making certain kinds of observations.  The fact that it does not describe social reality in its totality, or that it is not useful for all purposes, should not be surprising, since it is but one distinction among many.  Attempts to ignore or transcend this distinction end up reintroducing it covertly.  One must not transcend or replace distinctions, but rather, displace them, by means of other distinctions.  To distinguish, however, is inexorable.

 Anti-Functionalism

Giddens has called his theory an “anti-functionalist manifesto.  By “functionalism” he means the tendency to explain social phenomena in terms of their functional consequences for the social system.  Giddens objects to the use of functionalist language because it confuses the effects of such phenomena with their causes.  Only actors have reasons and motives, not social systems as aggregates of action.  In the language of functionalism, however, the unintended consequences generated by the intentional behavior of micro-level actors is explained instead as a result of the system, as a macro-level agent.

Although it is worthwhile pointing out that a discrepancy can exist between what people intend and what they achieve, the latter cannot, at the level of society, be a cause of itself.  As Rueschemeyer (1986) has argued, in order to be valid, functional explanations must show how the effects of a system bring about their own conditions through recursive feedback mechanisms.[2]  A social system cannot determine itself because it does not possess agency.   Structure, on the other hand, does possess some agency, but it can determine the range of possibility given at any moment in time.

Summary and Comments

 For Giddens, a pattern of relations (i.e. a "system") can be reconstituted across time, but he denies that the system itself does any constituting.  At the same time, individuals interacting with others do not create system(s), but can only reproduce and perhaps modify it to some extent.

Here is where I think Giddens’ theory needs clarification. 1)  First, the concept of system or structure should not be reified.  I understand Giddens as saying that there is a singular system that exists, or perhaps a range of such systems.  In contrast, I propose that social patterns are also models or hypotheses of social patterns.  They are not given, but require interpretation, or a reference frame.  The more reference frames available for the analysis, the more possible systems there are.  A macro, social-system, in other words, exists always as a micro, individual expectation (mental map or model).

This may seem like a minor point, but confusing our hypotheses of the big picture or the "totality" for the real "big picture" is a common mistake and grave error.  In a (second-order) model of models, which is capable of processing complexity (including mutually contradictory frames of reference), "system" is always "structurally coupled" to mental maps, in that a mind has to hypothesize a system.  This requires that the model include some people, exclude others, and consider some relations, while excluding others.  Social boundaries, however, are not physical boundaries, but boundaries of Meaning.   Importantly, the concept of system as unintended consequences of action, or macro-pattern, is not itself a system.  At this level of theorizing, moreover, social system is not specified.

Second, it is unclear how system reconstitutes itself across time.  What needs to be explained is how unintended consequences of people's actions "cause" or bring about actions that bring about those same unintended consequences.  Figure 80 is a model of Gidden’s model of the feedback relation between individually motivated actions (“actions”) and the unintended aggregate pattern of action (“system”).  Notice, however, that in Giddens model individually motivated actions reproduce the same unintended, aggregate pattern of action.  As a consequence, this is a circle rather than a spiral.  The time dimension is left out.  The aggregate pattern remains invariant and the actions variable.

Consider the example of the English language, which Giddens regards as the collective byproduct of speaking English.  Giddens does not specify, however, how can an outcome be a medium, and vice-versa.  A tidal wave does not create the medium of ocean water. In my opinion, English is not an invariant system in reality, but an invariant system held constant for analytical purposes by an observer.  The effects and consequences of action have to be theoretically specified.

Speaking can be specified in a number of different ways, and speaking can be classified according to a specification (or deviate from that specification to variable degrees), but the speaking does not generate the model according to which an observer classifies it.

Figure 80.  Giddens.  Feedback between action and system.

Figure 81.  Giddens Structuration theory.

Figure 82.  Giddens dynamic depiction of structuration.

To specify, consider the following schematic depiction of Giddens’ structuration model and compare it with a dynamic reinterpretation, both of which are reconstructed from Leydesdorff (2003).  Notice that Giddens neglects to specify the relation between system and structure.  In the dynamic version, structure is advanced ahead in the time axis to indicate that it is virtual and not present.   The static depiction of Giddens and the dynamic reformulation by Leydesdorff are presented in Figures 81 and 82.

Third, enablement and constraint are better and more precisely formulated as variation and selection.  If structure is not identical to what actually happened (historical instantiations) but remains broader, then the concept when theoretically specified would necessarily refer to actions that were possible but nevertheless not actualized, i.e. to de-selected cases.  The "structure" then is better understood as a phase space of possibilities for action.  The theoretical problem, however, is when Giddens, rather than viewing systems and actors themselves as the generators of the constraint part of the formula "enable and constrain" (that is, as the selectors of possibilities), equates the social structure with forces of constraint as well, which means that the structure is a subset of the phase space of possible actions, distinguished from other actions that are both a) not instantiated and b) not possibly instantiated at time t.  The difference is one between actions that are actually possible and those that are only theoretically possible, or perhaps, conceivable.  How does one distinguish between the two?  One then wonders, why not just include in the phase space, only those actions that are actually possible?  The model is incoherent:  structure refers to both a set of possible actions that is larger than those actions that are historically actualized, and is at the same time a constraint, that is, a subset of some unspecified superset of possible actions.

Fourth, Giddens implicitly equates "structure" with a range or distribution of possibilities, but does not distinguish these distributions for individual actors as opposed to multiple actors, nor does he explicitly acknowledge that structure can only be made theoretically meaningful in the context of a theory (entertained by a theorist/analyst) that posits the distribution as a selection from other possible distributions.  A distribution is by definition a variation, but from another perspective, we can compare distributions, and understand the distributions themselves as selections.  To be enabled means to have possibilities, or "choices", and to be constrained means that these choices are not infinite, and probably unequally weighted in terms of their likelihood.  Because Giddens' "social structure" is both enabling and constraining, it is by definition a distribution.  The two properties of variation (enablement) and constraint (selection), moreover, presuppose two alternative frames of reference:  in the former, a comparison is made between alternatives within the distribution, while in the latter a comparison is made between this distribution and other possible distributions, outside of it.

For example, when I speak English, I am instantiating, or selecting from certain words, but also producing a variation.  This can be viewed as a selection from within the distribution, specified for example by a dictionary, of all the possible English words constituting "English" (as specified by a hyper-reflexive observer).   Individual action, however, from this perspective, can only be understood as selection, or constraint, a subset.  It can be viewed as variation only from the perspective of the individual words.   In speaking English I am not reproducing the structure as set, but only the system as subset.  The structure has still to be specified as the unselected (but possible) cases. 

Furthermore, the 'phase space of possibilities' for actors taken individually isn't the same for the phase space of possibilities for systems of actors, taken as an aggregate.  Giddens wants to say that the actions of multiple individuals are structurated at the same time by the same structure, but are the possibilities specified for systems the same possibilities specified for individuals (subsets) of that system?  If not, then the structures are not identical.  In other words, they are different to the extent that we expect them to be.  There is no causal mechanism to explain the simultaneous transmission, or 'action at a distance' between individuals.  If the same structure exists to some extent "inside" different individuals at different places, how does this happen?  If we observe similarity between individuals, then mustn't this be explained via causal mechanisms?  The notion of a singular virtual structure structurating the actions of different individuals at the same time can only be done by an analyst who distinguishes his or her own observing from that of those being observed, and hence, places himself outside of, and external to the latent pattern manifested by those he observes.

How are these unselected, unrealized possibilities specified?  One could make reference to multiple systems across time, mining for historical examples in order to expand the range of the possible, but in doing so, one has also specified an instantiated historical system, albeit one with broader temporal parameters.  I do not generate English.  The "structure" or the expanded super-set of English requires analytical specification, a gestalt-switch, between seeing a communication as one thing (a subset) in comparison to other possible things, or many things (the set):  because meaning itself is the distinction between the actual and the possible, any meaningful statement will always be considered as a subset of other possible meaningful statements.  To say that I reproduce English as the super-set is incorrect. It is to confuse the historical instantiation (singular "action" or plural "system") with its meaning (i.e. with the counter-factual hypotheses as to what could have occurred, but didn't). 

In Giddens' model, system is the plural of action. To reformulate my original query, does

at-1 + bt-1 + ct-1 → Rt 

where Rt = "structure" and a, b, c designate motivated actions by different individuals? The answer is NO, because structure refers to the possible acts, that is, those actions not chosen or selected (i.e. not instantiated historically) as well as those that were and which can be reproduced.  Formalize this as:

Rt = (At-1+ Bt-1 + Ct-1)

where A, B, and C refer to the set of actual and possible actions.  The problem can be clarified by specifying a distinction between system and environment.   Ultimately, there is no outside to this model.  Society is a closed “system” consisting of action, structure, and system.

In my view, it is more productive to regard structure as a slowly changing parameter or condition of social communication that either may or may not be included in society as communication.   Whether or not this is part of society depends on whether society communicates about it.  The two media of communication and the constraints themselves, however, should be kept analytically distinct.

[1] Note:  for Giddens "action" cannot be reduced to individual acts:  "Action is not a combination of 'acts':  acts are constituted only by a discursive moment of attention to the duree of lived-through experience" (in Giddens Reader 1993:  90; from The Constitution of Society).

[2]  For example, in world systems analysis one might read that the semi-periphery exists because the world system “needs” it in order to diffuse tensions between the periphery and the core.  For WST, this raises the thorny issue of whether the triadic structure is merely an epiphenomenon or contingent by-product resulting from the activities of nation-states or whether the world-system acts as a supra-national agent in its own right to generate conditions that further its own reproduction.